I see that third_party/WebKit/LayoutTests/TestExpectations disables this test for Mac:

crbug.com/574283 [ Mac ] virtual/threaded/fast/scroll-behavior/first-scroll-runs-on-compositor.html [ Skip ]

This is probably unrelated, but just in case, let me CC people from issue 574283.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/f45a952f5bf8639ccd68886547f2d60a5de27c7c

commit f45a952f5bf8639ccd68886547f2d60a5de27c7c
Author: Lukasz Anforowicz <lukasza@chromium.org>
Date: Thu Oct 26 17:05:04 2017

Test exception for .../threaded/.../first-scroll-runs-on-compositor...

Bug:  778696 
No-Try: true
Change-Id: I9459651d78c33d475e2a03daaef1a13739507621
Reviewed-on: https://chromium-review.googlesource.com/739842
Reviewed-by: Alex Moshchuk <alexmos@chromium.org>
Commit-Queue: Łukasz Anforowicz <lukasza@chromium.org>
Cr-Commit-Position: refs/heads/master@{#511854}
[modify] https://crrev.com/f45a952f5bf8639ccd68886547f2d60a5de27c7c/third_party/WebKit/LayoutTests/FlagExpectations/site-per-process

Over to Kevin since this seems like an OOPIF issue. Kevin, could you take the initial look?

I see this test fail locally without --site-per-process as well. This test looks flaky: https://test-results.appspot.com/dashboards/flakiness_dashboard.html#testType=webkit_tests&tests=virtual%2Fthreaded%2Ffast%2Fscroll-behavior%2Ffirst-scroll-runs-on-compositor.html

I tried polling for the value of document.scrollingElement.scrollTop and in the cases where the test fails, the value was 524 instead of the expected 500. It's the same issue with and without --site-per-process.

Ah, ok, thanks Kevin. In that case, over to sunyunjia who owns smooth scrolling.

This is happening a lot on the linux_chromium_rel_ng tryserver, causing many CLs to be retried. If the test is flaky then it should be disabled.

https://ci.chromium.org/buildbot/tryserver.chromium.linux/linux_chromium_rel_ng/?limit=200

Just a few failures found:
https://ci.chromium.org/buildbot/tryserver.chromium.linux/linux_chromium_rel_ng/586764
https://ci.chromium.org/buildbot/tryserver.chromium.linux/linux_chromium_rel_ng/586749
https://ci.chromium.org/buildbot/tryserver.chromium.linux/linux_chromium_rel_ng/586728
https://ci.chromium.org/buildbot/tryserver.chromium.linux/linux_chromium_rel_ng/586727
https://ci.chromium.org/buildbot/tryserver.chromium.linux/linux_chromium_rel_ng/586722

One failure's archived test results:

https://storage.googleapis.com/chromium-layout-test-archives/linux_chromium_rel_ng/586722/layout-test-results/results.html

Upgrading to P1 -- this needs to be addressed ASAP.

Haven't seen this test fail in the last 200 runs of linux_chromium_rel_ng (from 589982  to 590192). 
I believe no more sheriff involvement is necessary, so removing Sheriff-Chromium label.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/8e638519a02830035b58a02dd23dc5bb86488d34

commit 8e638519a02830035b58a02dd23dc5bb86488d34
Author: Sandra Sun <sunyunjia@chromium.org>
Date: Fri Nov 17 15:37:49 2017

Disable first-scroll-runs-on-compositor.html

virtual/threaded/fast/scroll-behavior/first-scroll-runs-on-compositor.html
is flaky.

This patch temporarily disables the test, and a following test
will update the test and enable it back again.

Bug:  778696 
Change-Id: Id27ccb14e0deda20e46478a237ed07962ad7eb5d
Reviewed-on: https://chromium-review.googlesource.com/770599
Reviewed-by: Philip Jägenstedt <foolip@chromium.org>
Commit-Queue: Philip Jägenstedt <foolip@chromium.org>
Cr-Commit-Position: refs/heads/master@{#517401}
[modify] https://crrev.com/8e638519a02830035b58a02dd23dc5bb86488d34/third_party/WebKit/LayoutTests/TestExpectations

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/c9f77bf3e3b255910d644c32fca71b960273865b

commit c9f77bf3e3b255910d644c32fca71b960273865b
Author: Sandra Sun <sunyunjia@chromium.org>
Date: Tue Nov 21 16:33:37 2017

Rewrite first-scroll-runs-on-compositor for more robustness.

Currently, first-scroll-runs-on-compositor.html uses ShouldBecomeEqual()
from js-test.js to check the final scroll destination. However, it
checks after 0.5secs, which may not be enough for all the platforms,
making the test flaky.

This patch rewrites the test using testharness.js and only checkes after
the page no longer scrolls or has reached the specified destination,
which would be independent from the platform's performance, making the
test robust.

Bug:  778696 
Change-Id: Icb91755a3eaecbb5a76023a303fcf9a27f713665
Reviewed-on: https://chromium-review.googlesource.com/777606
Commit-Queue: Sandra Sun <sunyunjia@chromium.org>
Reviewed-by: David Bokan <bokan@chromium.org>
Cr-Commit-Position: refs/heads/master@{#518273}
[modify] https://crrev.com/c9f77bf3e3b255910d644c32fca71b960273865b/third_party/WebKit/LayoutTests/FlagExpectations/enable-browser-side-navigation
[modify] https://crrev.com/c9f77bf3e3b255910d644c32fca71b960273865b/third_party/WebKit/LayoutTests/FlagExpectations/site-per-process
[modify] https://crrev.com/c9f77bf3e3b255910d644c32fca71b960273865b/third_party/WebKit/LayoutTests/TestExpectations
[delete] https://crrev.com/65ae4d02fdbe74f2e1bd708edd1beef53f9841b4/third_party/WebKit/LayoutTests/virtual/threaded/fast/scroll-behavior/first-scroll-runs-on-compositor-expected.txt
[modify] https://crrev.com/c9f77bf3e3b255910d644c32fca71b960273865b/third_party/WebKit/LayoutTests/virtual/threaded/fast/scroll-behavior/first-scroll-runs-on-compositor.html

Making fixed since https://test-results.appspot.com/dashboards/flakiness_dashboard.html#testType=webkit_tests&tests=first-scroll-runs-on-compositor.html shows all green. Thanks!